The Pentagon’s central research arm said Thursday it plans to focus on less complex military systems that hold the potential to provide major technological breakthroughs capable of ensuring U.S. military superiority on the battlefields of the future — including cyberspace.
Arati Prabhakar, director of the Defense Advanced Research Projects Agency, told members of the House Armed Services Committee that the technologies once considered to be in the exclusive realm of the U.S. military are rapidly finding their way to the militaries of competitor nations and potential adversaries.
“The rest of the world doesn’t stand still,” Prabhakar said, during a hearing following the agency’s release of its biennial report, “Breakthrough Technologies for National Security.” “Today we see other capabilities developing around the world that put our advantage at risk.”
Among the most important projects designed to reverse this trend is DARPA’s work to control the electromagnetic spectrum. “This is new work at the level of devices, new systems architectures, new algorithms, new manufacturing technologies, all of which together I think can give us a chance to move into a future…where we can control the electromagnetic spectrum in real time,” Prabhakar said.
In an effort to reduce the complexity of military systems, DARPA is shifting its focus on more modular technologies and systems that do not rely exclusively on GPS satellites for navigation and timing.
Big data is also a big focus of the report. “Increasingly, the bottleneck to wise decision making is not a lack of data but a lack of capacity to identify and understand the most important data,” the report states. “Toward this end, DARPA is developing novel approaches to deriving insights from massive datasets and to mapping behavior patterns at scale, including algorithms to quickly identify anomalous threat-related behaviors of systems, individuals and groups.”
Prabhakar highlighted a program started last year called Memex, which is quickly revolutionizing the world of public Internet search beyond the capabilities of even Google. Although the program started as an effort to combat human trafficking, it is now being used to fight the online activities of the self-proclaimed Islamic State.
Prabhakar described Memex as “deep domain-specific search” targeting public websites. “It lets [analysts] see websites out beyond those that are indexed by commercial search engines like Google of Bing. And then secondly this tool automatically maps patterns, linkages and relationships across vast numbers of websites,” Prabhakar said.
Along those lines, artificial intelligence — or machine learning — is also playing a major role in some of the potential breakthrough programs underway at DARPA. The agency has started initiatives focused on enabling computers to read and understand massive amounts of scientific journal information and visual media reasoning, which will enable analysts to conduct search queries against photo content.
DARPA’s work on machine learning could also one day eliminate the need for expert coders to produce intelligence applications. DARPA’s Probabilistic Programming for Advancing Machine Learning (PPAML) program aims to create “user-friendly programming languages to simplify and democratize the now-arcane art of building machine-learning applications,” the report states. “Doing so would accelerate the development of intelligence-intensive applications, from email spam filters to smartphone personal assistants to self-driving vehicles, by obviating the need for hard-to-find experts to build custom software from scratch.”